Method and system for automatically identifying and processing abnormity of shield guiding system

Abstract

The invention discloses a method and a system for automatically identifying and processing abnormity of a shield guide system, relates to the technical field of shield machine control, and solves the technical problem that in the prior art, when measured attitude data of the guide system are abnormal due to unforeseen circumstances, tunneling construction cannot be normally carried out. The method comprises the following steps: training a neural network model by using historical shield construction data to obtain an identification model; during shield construction, shield construction data are updated in real time, the real-time shield data are input into the recognition model to obtain a posture data prediction result, and if the difference value between the data of a certain posture data after the specified period of the corresponding curve segment type and the posture data prediction result of the recognition model exceeds a threshold value, it is judged that the guiding system is abnormal; if the guiding system is judged to be abnormal, taking a difference value between an ideal predicted value and an actual predicted value of the attitude deviation of the guiding system at the current position as an estimated value of attitude deviation data, and performing tunneling control based on the estimated value; the shield tunneling construction quality is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of shield machine control, and in particular to a method and system for automatically identifying and processing abnormalities of a shield guide system. Background Art

[0002] The shield machine, whose full name is tunnel boring machine, is a special engineering machinery for tunnel excavation. Modern shield tunnel boring machines integrate optical, mechanical, electrical, hydraulic, sensor and information technology. They have the functions of excavating and cutting soil, transporting soil and slag, assembling tunnel lining, measuring and guiding, and correcting deviations. They involve multiple disciplines and technologies such as geology, civil engineering, machinery, mechanics, hydraulics, electricity, control, and measurement. They have been widely used in tunnel projects such as subways, railways, highways, municipal engineering, and hydropower.

[0003] During tunnel excavation, the shield machine needs to control its position and direction in a timely manner. The automatic measurement and guidance system continuously measures the shield machine's attitude data and displays it on the computer screen in the main control room. The driver adjusts the shield machine's excavation parameters according to the attitude data to ensure that the shield machine accurately excavates according to the tunnel design axis (DTA).

[0004] When the laser beam of the automatic measurement and guidance system is blocked by foreign objects, or when the total station or laser target malfunctions, the posture data measured by the guidance system will become inaccurate or out of date, affecting the normal progress of the excavation construction. Summary of the invention

[0005] In order to solve the problems existing in the above-mentioned prior art, the present invention provides a method and system for automatically identifying and processing abnormalities of a shield guide system, thereby solving the technical problem that the prior art cannot carry out excavation construction normally when the measured posture data of the guide system is abnormal due to unexpected circumstances.

[0006] A method for automatically identifying and processing anomalies of a shield guidance system, comprising:

[0007] A neural network model is trained using historical shield construction data to obtain a recognition model. The historical shield construction data includes interval data, target parameters and corresponding target data as outputs of the recognition model, and related parameters and corresponding related data as inputs of the recognition model. The target data and related data are divided into three types of data: straight line segment data D1, circular curve segment data D2, and transition curve segment data D3 using the interval data. Specifically, the target parameters and related parameters are specific data types, and the target data and related data are data contents corresponding to the data types.

[0008] During shield construction, the shield construction data is updated in real time, and the real-time shield data is input into the recognition model to obtain the posture data prediction result. If the difference between the data of a certain posture data after the specified period of the corresponding curve segment type and the posture data prediction result of the recognition model exceeds the threshold, the guidance system is judged to be abnormal;

[0009] If the guidance system is judged to be abnormal, the expected stroke difference of the current position is calculated and input into the recognition model to obtain the ideal predicted value of the guidance system attitude deviation at the current position. At the same time, the actual stroke difference of the current position is input into the recognition model to obtain the actual predicted value of the guidance system attitude deviation at the current position. The difference between the two is used as the estimated value of the attitude deviation data when the guidance system is abnormal, and excavation control is performed based on the estimated value.

[0010] Furthermore, three neural network models are trained using the three types of data D1, D2, and D3 to obtain three corresponding recognition models.

[0011] Furthermore, the method for obtaining the specified period of the corresponding curve segment type is as follows: count the time required for each update of the three types of data D1, D2, and D3, calculate the mean and standard deviation of the counted time data, and take the sum of the mean and the standard deviation as the maximum effective update period of the guidance system data in the corresponding interval, that is, the specified period T1, T2, and T3.

[0012] Furthermore, the target parameter data is shield body posture data, including the horizontal offset and vertical offset of the center points of the shield head, shield middle and shield tail automatically measured by the guidance system as target parameters, and the horizontal and vertical trend deviations of the front and rear shields; the related parameters include the strokes of four groups of propulsion cylinders and the displacements of four groups of articulated cylinders, and the four groups include an upper group, a lower group, a left group and a right group.

[0013] Further, the calculation of the expected stroke difference of the current position includes: calculating the expected value of the change in the propulsion cylinder stroke difference within the current loop according to the interval data, and then constructing the objective function of the propulsion cylinder stroke difference within the current loop with respect to the cylinder stroke of the current position.

[0014] Further, the calculation of the expected value of the change in the current propulsion cylinder stroke difference in the ring according to the interval data includes:

[0015] For the data of the straight line segment, the expected value Y1 of the change in the stroke difference between the left and right groups of propulsion cylinders and the change in the stroke difference between the upper and lower groups of propulsion cylinders is set to zero;

[0016] For the data of the circular curve segment, the expected value Y2 of the change in the stroke difference of the propulsion cylinder in the current ring is calculated based on the interval data segment ring width l, segment outer diameter r, and turning radius R;

[0017]

[0018] For the data of the transition curve segment, the expected value Y3 of the change in the stroke difference of the propulsion cylinder in the current ring is calculated based on the expected value Y′ of the stroke difference of the circular curve segments immediately before and after, and the number of rings h contained in the interval;

[0019]

[0020] Further, the objective function of constructing the propulsion cylinder stroke difference in the current loop with respect to the cylinder stroke at the current position includes: determining the expected value of the stroke difference at the current moment f(x) according to the formula:

[0021]

[0022] Wherein, Y can be Y1, Y2, or Y3 according to the type of curve segment corresponding to the current position.

[0023] Furthermore, when the shield tunneling is manually controlled, the tunneling control can be performed through the estimated value of the posture deviation; when the shield tunneling is automatically controlled, the tunneling control can also be performed directly based on the error between the expected stroke difference and the actual stroke difference at the current position.

[0024] Furthermore, the neural network model contains two hidden layers.

[0025] A system for automatically identifying and processing anomalies of a shield guidance system, used to implement a method for automatically identifying and processing anomalies of a shield guidance system, comprising a data acquisition module, an anomaly identification module and an anomaly processing module;

[0026] The data acquisition module collects and updates shield data in real time;

[0027] The abnormality recognition module includes a recognition model, and the real-time shield data is input into the recognition model to obtain the posture data prediction result. If the difference between the data of a certain posture data after the specified period of the corresponding curve segment type and the posture data prediction result of the recognition model exceeds a threshold, it is judged that the guidance system is abnormal;

[0028] The exception handling module is used to calculate the expected stroke difference at the current position and input it into the recognition model to obtain the ideal predicted value of the attitude deviation of the guidance system at the current position when the guidance system is abnormal. At the same time, the actual stroke difference at the current position is input into the recognition model to obtain the actual predicted value of the attitude deviation of the guidance system at the current position. The difference between the two is used as the estimated value of the attitude deviation data when the guidance system is abnormal, and excavation control is performed based on the estimated value.

[0029] The beneficial effects of the present invention include:

[0030] The present invention can judge the working state of the guide system during the tunneling process of the shield machine, and provide relatively reliable posture data when the guide system is abnormal, so as to ensure that the shield machine can continue to tunnel, thereby improving the quality of shield tunneling construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a flowchart of a method for automatically identifying and processing anomalies in a shield guidance system according to an embodiment of the present application.

[0032] Figure 2 Schematic diagram of the structure of the prediction deviation neural network involved in the embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0034] A method for automatically identifying and processing abnormalities in a shield guidance system, such as Figure 1 As shown, including:

[0035] The neural network model is trained by using historical shield construction data to obtain a recognition model, wherein the historical shield construction data includes interval data, target parameters and corresponding target data as outputs of the recognition model, and related parameters and corresponding related data as inputs of the recognition model. The target data and related data are divided into three types of data: straight line segment data D1, circular curve segment data D2, and transition curve segment data D3 by using the interval data;

[0036] Specifically, historical construction data is collected to obtain historical shield construction data, including obtaining original shield tunneling data, interval data, and fault data from the historical construction data, and selecting target parameters that need to be identified and related parameters for model training from the original shield tunneling data;

[0037] According to the fault data, the data of the shield machine during normal operation is filtered out from the original shield tunneling data, and then the target data corresponding to the target parameters and the related data corresponding to the related parameters are obtained from the filtered data. Finally, the obtained target data and related data are divided into three types of data in different tunneling intervals according to the interval data; specifically, the target parameters and related parameters are specific data types, and the target data and related data are the data contents corresponding to the data types. Specific method:

[0038] According to the advancing speed, delete the data of zero speed and the first 10% of the non-zero speed segment in each excavation ring to obtain the data when the shield machine is advancing steadily;

[0039] Filter the data of shield machine during normal operation according to the occurrence time and recovery time of various faults in the fault data;

[0040] Target data corresponding to the target parameters and related data corresponding to the related parameters are obtained from the filtered data.

[0041] The acquired target data and related data are divided into three types of data: straight line segment data D1, circular curve segment data D2, and transition curve segment data D3 according to the interval data.

[0042] During shield construction, the shield construction data is updated in real time, and the real-time shield data is input into the recognition model to obtain the posture data prediction result. If the difference between the data of a certain posture data after the specified period of the corresponding curve segment type and the posture data prediction result of the recognition model exceeds the threshold, the guidance system is judged to be abnormal;

[0043] Specifically, the data in the shield construction process is updated in real time, the changes in the posture data of each guide system are detected, and the posture data of the guide system is predicted through the corresponding type of neural network model according to the changes in the propulsion cylinder stroke and the articulation cylinder stroke. If the difference between the data after a certain posture data exceeds the maximum effective update cycle of the corresponding curve segment and the data predicted by the neural network model exceeds the set value, it is judged that the guide system is abnormal, and the set value is 30% of the neural network prediction value;

[0044] If the guidance system is judged to be abnormal, the expected stroke difference of the current position is calculated and input into the recognition model to obtain the ideal predicted value of the guidance system attitude deviation at the current position. At the same time, the actual stroke difference of the current position is input into the recognition model to obtain the actual predicted value of the guidance system attitude deviation at the current position. The difference between the two is used as the estimated value of the attitude deviation data when the guidance system is abnormal, and excavation control is performed based on the estimated value.

[0045] In another embodiment, the target parameter data is shield body posture data, including the horizontal offset and vertical offset of the center points of the shield head, shield middle and shield tail automatically measured by the guidance system as target parameters, and the horizontal and vertical trend deviations of the front and rear shields; the relevant parameter data include the strokes of four groups of propulsion cylinders and the displacements of four groups of articulated cylinders, and the four groups include an upper group, a lower group, a left group and a right group.

[0046] In another embodiment, three neural network models are trained using the three types of data D1, D2, and D3 to obtain three corresponding recognition models. Specifically, the specific method of training the BP neural network model is: the four parameters of the target parameter data of the shield head horizontal deviation, shield head vertical deviation, shield head horizontal deviation, and shield head vertical deviation in the three types of data D1, D2, and D3 are used as the output of the BP neural network model, and the horizontal and vertical cylinder stroke differences formed by the cylinder strokes of the four groups of propulsion cylinders of the relevant parameter data A, B, C, and D and the horizontal and vertical cylinder displacement differences formed by the cylinder displacements of the four groups of articulated cylinders of the upper right, lower right, lower left, and upper left are used as inputs to train three neural network models with two hidden layers respectively, such as Figure 2 As shown, a mapping model from relevant parameters to target parameters is obtained.

[0047] In another embodiment, the method for obtaining the specified period of the corresponding curve segment type is as follows: the time required for each update of the three types of data D1, D2, and D3 is counted respectively, the mean and standard deviation of the counted time data are calculated, and the sum of the mean and the standard deviation is taken as the maximum effective update period of the guidance system data in the corresponding interval, that is, the specified period T1, T2, and T3.

[0048] In another embodiment, the calculation of the expected stroke difference of the current position includes: calculating the expected value of the change in the propulsion cylinder stroke difference within the current loop based on the interval data, and then constructing the objective function of the propulsion cylinder stroke difference within the current loop with respect to the cylinder stroke at the current position.

[0049] In another embodiment, the expected value of the change amount of the propulsion cylinder stroke difference in the current ring calculated according to the interval data (calculated by the measured interval data, indicating the change amount of the cylinder stroke difference before and after completing the propulsion task of the current ring) includes:

[0050] For the data of the straight line segment, the expected value Y1 of the change in the stroke difference between the left and right groups of propulsion cylinders and the change in the stroke difference between the upper and lower groups of propulsion cylinders is set to zero;

[0051] For the data of the circular curve segment, the expected value Y2 of the change in the stroke difference of the propulsion cylinder in the current ring is calculated based on the interval data segment ring width l, segment outer diameter r, and turning radius R;

[0052]

[0053] For the data of the transition curve segment, the expected value Y3 of the change in the stroke difference of the propulsion cylinder in the current ring is calculated based on the expected value Y′ of the stroke difference of the circular curve segments immediately before and after, and the number of rings h contained in the interval;

[0054]

[0055] In another embodiment, the objective function of constructing the current ring propulsion cylinder stroke difference with respect to the current position cylinder stroke includes: determining the current moment stroke difference expected value f(x) according to the formula based on the expected value Y of the current ring propulsion cylinder stroke difference change, the current ring expected excavation progress X (pre-planned by the construction team and included in the interval data), the current moment excavation progress x (measured in real time during the shield machine advancement process), and the current ring initial stroke difference Y0 (measured and obtained before the current ring is advanced)

[0056]

[0057] Wherein, Y can be Y1, Y2, or Y3 according to the type of curve segment corresponding to the current position.

[0058] In another embodiment, when the shield tunneling is manually controlled, the tunneling control can be performed by the estimated value of the posture deviation; when the shield tunneling is automatically controlled, the tunneling control can also be performed directly based on the error between the expected stroke difference and the actual stroke difference at the current position.

[0059] Specifically, the expected travel difference f(x0) of the current position x0 is input into the neural network model to obtain the ideal predicted value E of the guidance system attitude deviation at the current position. p , which is used as the offset of the model prediction;

[0060] The actual travel difference g(x0) at the current position is input into the neural network model to obtain the actual predicted value E′ of the guidance system attitude deviation at the current position p According to the formula

[0061] E=E′ p -E p

[0062] Obtain an estimated value E of the attitude deviation data when the guidance system is in an abnormal situation.

[0063] When the shield tunneling is manually controlled, the tunneling control can be carried out through the estimated value E of the posture deviation; when the shield tunneling is automatically controlled, the tunneling control can also be carried out directly based on the error between the expected stroke difference f(x0) and the actual stroke difference g(x0) of the current position x0.

[0064] A system for automatically identifying and processing anomalies of a shield guidance system, used to implement a method for automatically identifying and processing anomalies of a shield guidance system, comprising a data acquisition module, an anomaly identification module and an anomaly processing module;

[0065] The data acquisition module collects and updates shield data in real time;

[0066] The abnormality recognition module includes a recognition model, and the real-time shield data is input into the recognition model to obtain the posture data prediction result. If the difference between the data of a certain posture data after a specified period and the posture data prediction result of the recognition model exceeds a threshold, it is judged that the guidance system is abnormal;

[0067] The exception handling module is used to calculate the expected stroke difference at the current position and input it into the recognition model to obtain the ideal predicted value of the attitude deviation of the guidance system at the current position when the guidance system is abnormal. At the same time, the actual stroke difference at the current position is input into the recognition model to obtain the actual predicted value of the attitude deviation of the guidance system at the current position. The difference between the two is used as the estimated value of the attitude deviation data when the guidance system is abnormal, and excavation control is performed based on the estimated value.

[0068] The above-mentioned embodiments only express the specific implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the protection scope of the present application. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the technical solution concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. A method for automatically identifying and processing abnormalities in a shield guidance system, characterized in that: include: The neural network model is trained by using historical shield construction data to obtain a recognition model, wherein the historical shield construction data includes interval data, target parameters and corresponding target data as outputs of the recognition model, and related parameters and corresponding related data as inputs of the recognition model. The target data and related data are divided into three types of data: straight line segment data D1, circular curve segment data D2, and transition curve segment data D3 by using the interval data; During shield construction, the shield construction data is updated in real time, and the real-time shield data is input into the recognition model to obtain the posture data prediction result. If the difference between the data of a certain posture data after the specified period of the corresponding curve segment type and the posture data prediction result of the recognition model exceeds the threshold, the guidance system is judged to be abnormal; If the guidance system is judged to be abnormal, the expected stroke difference of the current position is calculated and input into the recognition model to obtain the ideal predicted value of the guidance system attitude deviation at the current position. At the same time, the actual stroke difference of the current position is input into the recognition model to obtain the actual predicted value of the guidance system attitude deviation at the current position. The difference between the two is used as the estimated value of the attitude deviation data when the guidance system is abnormal, and excavation control is performed based on the estimated value.

2. A method for automatically identifying and processing shield guidance system anomalies according to claim 1, characterized in that: Three neural network models are trained using the three types of data D1, D2, and D3 to obtain three corresponding recognition models.

3. The method for automatically identifying and processing abnormalities of a shield guidance system according to claim 1 is characterized in that: The method for obtaining the specified period of the corresponding curve segment type is as follows: respectively count the time required for each update of the three types of data D1, D2, and D3, calculate the mean and standard deviation of the counted time data, and take the sum of the mean and the standard deviation as the maximum effective update period of the guidance system data in the corresponding interval, that is, the specified period T1, T2, and T3.

4. The method for automatically identifying and processing anomalies of a shield guidance system according to claim 1 is characterized in that: The target parameters are the horizontal offset and vertical offset of the center points of the shield head, shield middle and shield tail, and the horizontal and vertical deviations of the front and rear shields automatically measured by the guidance system; the related parameters include the strokes of four groups of propulsion cylinders and the displacements of four groups of articulated cylinders, and the four groups include an upper group, a lower group, a left group and a right group.

5. The method for automatically identifying and processing abnormalities of a shield guidance system according to claim 1 is characterized in that: The calculation of the expected stroke difference of the current position includes: calculating the expected value of the change in the propulsion cylinder stroke difference in the current loop according to the interval data, and then constructing the objective function of the propulsion cylinder stroke difference in the current loop with respect to the cylinder stroke at the current position.

6. A method for automatically identifying and processing anomalies of a shield guidance system according to claim 5, characterized in that: The calculation of the expected value of the change in the current propulsion cylinder stroke difference in the ring according to the interval data includes: For the data of the straight line segment, the expected value Y1 of the change in the stroke difference between the left and right groups of propulsion cylinders and the change in the stroke difference between the upper and lower groups of propulsion cylinders is set to zero; For the data of the circular curve segment, the expected value Y2 of the change in the stroke difference of the propulsion cylinder in the current ring is calculated based on the interval data segment ring width l, segment outer diameter r, and turning radius R; For the data of the transition curve segment, the expected value Y3 of the change in the stroke difference of the propulsion cylinder in the current ring is calculated based on the expected value Y′ of the stroke difference of the circular curve segments immediately before and after, and the number of rings h contained in the interval; 7. A method for automatically identifying and processing abnormalities of a shield guidance system according to claim 6, characterized in that: The objective function of constructing the current loop propulsion cylinder stroke difference with respect to the current position cylinder stroke includes: determining the current moment stroke difference expected value f(x) according to the formula: Wherein, Y can be Y1, Y2, or Y3 according to the type of curve segment corresponding to the current position.

8. The method for automatically identifying and processing anomalies of a shield guidance system according to claim 1 is characterized in that: When the shield tunneling is controlled manually, the tunneling control can be carried out through the estimated value of the posture deviation; when the shield tunneling is controlled automatically, the tunneling control can also be carried out directly based on the error between the expected stroke difference and the actual stroke difference at the current position.

9. The method for automatically identifying and processing abnormalities of a shield guidance system according to claim 1, characterized in that: The neural network model contains two hidden layers.

10. A system for automatically identifying and processing anomalies of a shield guidance system, used to implement a method for automatically identifying and processing anomalies of a shield guidance system as described in any one of claims 1 to 9, comprising a data acquisition module, an anomaly identification module and an anomaly processing module; The data acquisition module collects and updates shield data in real time; The abnormality recognition module includes a recognition model, and the real-time shield data is input into the recognition model to obtain the posture data prediction result. If the difference between the data of a certain posture data after the specified period of the corresponding curve segment type and the posture data prediction result of the recognition model exceeds a threshold, it is judged that the guidance system is abnormal; The exception handling module is used to calculate the expected stroke difference at the current position and input it into the recognition model to obtain the ideal predicted value of the guidance system posture deviation at the current position when the guidance system is abnormal. At the same time, the actual stroke difference at the current position is input into the recognition model to obtain the actual predicted value of the guidance system posture deviation at the current position. The difference between the two is used as the estimated value of the posture deviation data when the guidance system is abnormal, and excavation control is performed based on the estimated value.

Images